AI agents spark an identity cleanup

Identity and access management is receiving renewed attention as AI agents transition from experimental pilots to production use. The rapid deployment of these AI agents is revealing the legacy identity tools, integration challenges, and security vulnerabilities that enterprises have accumulated over the years. Security professionals are feeling pressure to enable rapid business growth through AI while maintaining control over the agents' access to sensitive data.

Jon Addison, Okta's chief revenue officer, explained that customers view agentic security as an extension of their existing identity work, rather than a completely new problem. Addison stated, "The power of this technology is enormous," and emphasized that businesses are under significant pressure to innovate quickly. However, they recognize that achieving this innovation requires visibility and control, which aligns closely with the identity management solutions they have been developing.

Many organizations have historically constructed their identity stacks incrementally, leading to a tangled network of point solutions that require ongoing maintenance and integration efforts. As AI agents demand more extensive access, Addison noted that these legacy systems can introduce vulnerabilities, prompting customers to modernize and consolidate their identity infrastructure before scaling AI deployments.

Some companies are proactively establishing reference architectures to guide their AI agent initiatives, while others are struggling to implement security controls for already-existing "shadow" agents in production. Addison highlighted that the focus is on improving efficiency and augmenting human capabilities, but anticipates future challenges related to the increasing autonomy of AI agents and the need for novel approaches to managing their scale.